Drunk statistician wastes time

[kiwistiller]

So we had a few drinks after work today and I'm fairly sure I'm not good to drive. I need to fill in an hour. So, I thought I'd detail a theory I've been pondering (just in the name of a laugh, hope no one takes this seriously).

I was thinking, right, that we should index LWTCS’s use of punctuation. I have a theory that it relates to alcohol consumption. So, if Np is the number of ,’s or .’s in a post (N, + N.), and Nt is the total sentences in the post, then we should be able to extract a punctuation index. So, Pi (punctuation index) will = (Np/Nt) x 100 So, if LWTCS is more or less sober, we can expect 1/1 x 100 = 100. This is our base rate. Lets call it one drink.

Now, lets say LWTCS has a few drinks, and becomes a little punctuation happy. For example, in this thread http://homedistiller.org/forum/viewtopi ... 5#p6805745 we can see that there are 30 punctuation marks in 4 sentences. This gives us 30/4 x 100 = 750, and if the rate of punctuation use is directly and proportionately related to the amount of drinks consumed, we can assume that LWTCS has probably consumed about 7 or 8 drinks (drink defined as 1 standard drink, as defined in the civilised world (nz and oz) as 10g or 12.7ml of pure ethanol).

To assess the validity of this claim, we need to run a controlled experiment whereby we feed LWTCS measured amounts booze and make him type. We can then compare the observed punctuation usage with the predictions of the index as used in a regressive manner, and gather a set of residuals. Using the sum of squared residuals, we can find a probability curve which has a Students t-distribution with N-1 degrees of freedom, where N is the number of times we give him drinks and make him type (call it observations). Using the t-distribution, we can find a confidence interval that should prove the above relationship statistically significant.

LWTCS? Your thoughts?

[olddog]

Ha Ha Ha I like that

[HookLine]

Correlation is not causation.

[kiwistiller]

Correlation is not causation.

true, but if we do a paired sample experiment with a set of control observations at similar time intervals in a controlled environment, we could show causation. tho either way its a bit dubious with a sample of one. but I can't think of anyone else that uses lots of punctuation when drunk except for myself, and as I subscribe to the positivism philosophy rather than the interpretist, I cannot, as the researcher, include myself as part of the study.

[rubber duck]

HMM that's funny i think but, I'll have to run it by a few of my stats friends to be shure lol. Ether way you just messed up my buzz.

Im going to go fix that now thanks for the exuse.

[kiwistiller]

make sure you get the stats friends drunk first, or it won't make sense to them, and they'll see that its just bullshit.

[rubber duck]

They all do psyc stats so I'm not shure if they know what bs is. Any way i just had 2 more drinks and now this post is realy funny. I think ill have 3 more and read it agian.

[HookLine]

Correlation is not causation.

true, but if we do a paired sample experiment with a set of control observations at similar time intervals in a controlled environment, we could show causation. tho either way its a bit dubious with a sample of one. but I can't think of anyone else that uses lots of punctuation when drunk except for myself, and as I subscribe to the positivism philosophy rather than the interpretist, I cannot, as the researcher, include myself as part of the study.

Whoodywatsit?

Just run a standard RCT.

tho either way its a bit dubious with a sample of one.

Well, I didn't want to point out such an obvious flaw. Clearly a proper RCT will need a lot more test subjects. I mean, we can't have such important questions going unanswered. I volunteer for the sacrifice, as long as I am not in the control group.

[kiwistiller]

tho either way its a bit dubious with a sample of one.

Well, I didn't want to point out such an obvious flaw. Clearly a proper RCT will need a lot more test subjects. I mean, we can't have such important questions going unanswered. I volunteer for the sacrifice, as long as I am not in the control group.

I'm not really convinced we need more participants, or to bother with a full experimental design. After all, no one else seems to exhibit what I'm calling 'anomalous punctuation syndrome', and we have plenty of base rate observations (thousands and thousands) here on this forum. I think you have our objectives muddled. I don't want to prove that drinking causes punctuation, I want to prove that LT is strange, different from the rest of us. a RMD should be good for this, and in fact desirable, one of the key uses of repeated measures is to enable statistical inference from small samples.

[plonker]

This is a great thread! almost fell off my chair laughing..

Where Larry?? we need to test this!

*grin*

[olddog]

Larrys asleep now unless he's doing night shift

[LWTCS]

Thats funny as hell Kiwi.

And I recon (to some degree) true.

This may require more testing...............(I mean..)

[LWTCS]

BTW,

7 or 8 drinks at what proof?
I don't really prefer 40% abv anymore.

I'm usually 2 or 3 at 65%

[kiwistiller]

snip
(drink defined as 1 standard drink, as defined in the civilised world (nz and oz) as 10g or 12.7ml of pure ethanol).

[HookLine]

I'm not really convinced we need more participants, or to bother with a full experimental design. After all, no one else seems to exhibit what I'm calling 'anomalous punctuation syndrome', and we have plenty of base rate observations (thousands and thousands) here on this forum.

Uncontrolled for alcohol consumption, use of (freely available) anomalous punctuation checking software, shoe size, etc. All important potential confounders. Gotta do this rigourously. No sloppy science here, dude.

I think you have our objectives muddled.

Do I? My objective is to make sure the study is long running, has an active 'treatment' arm, and I get into it.

Ah, the nobility of pure science.

[kiwistiller]

alright, you've talked me into it, we'd better use a longitudinal design. how about you, me, larry, and lets say hawke (coz I reckon he'd be fun OTP) can be the treatment group, and everyone else can be control.

EDIT none of this double blind bullsh*t, here, I'm damn sure not going in the control.

[HookLine]

Gee, that was hard.

And I agree completely about the double blinding. Clearly unnecessary in this specialised study. Will just confuse the issue.

Also, let's call it a large pilot study, that should get it past the ethics committee.

[kiwistiller]

Also, let's call it a large pilot study, that should get it past the ethics committee.

I was just going to put in a 'low risk' application in the hope they rubber stamp it without really paying attention

[HookLine]

You've done this before, haven't you.

[goinbroke2]

I'D VOLUTEER TOO BUT NORMLY IM TOO LIKKERD WHEN i DO COME ON HERE AND ITS LAET AT NIGHT AND THE OLD LADY IS MAD AS A HEN IN A HORHOUSE...NO WAIT THATS NOT RIGHT..OH CRAP i'M YELLING1!!1!

hey guess what I did? I invented something! I put 2 peppercorns in ech 750 bottle and holy crap it burns your lips! It's like a warm fuzzy feeling but only 40%abv! Oh yeah, the reson I can't be in the group is becuse the next day I lolk at wat I wrote and get embareassed. some of it is funy but more is "what was i talkimng about? wow,7;26 friday night and im done! all she wrote, likkered, smashed, firkin legless, plastered good night




i love my still!

[kiwistiller]

coughcough19/12*100=punctuation index = 158 cough cough.

[cob]

there is a civilized world AND an ethics committee ? holy sh!!! cob

[goinbroke2]

Hmmm, searching I find I was on here last night too.
Not as bad as a couple dragrace forums I was on last night. I gotta stay away from the computor when I've been drinking.